The frequency of regular use of Vitamin A preparations was compared among 800 newly diagnosed cancer patients and 3433 patients with certain nonmalignant conditions. Overall there was no convincing evidence that regular consumption of such preparations protected against the development of cancer (risk ratio 0.85). However, there was a suggestion of a negative association among men (risk ratio 0.54, p less than 0.01) but not among women (risk ratio 1.11). The negative association among men did not increase with the period that such preparations had been regularly taken and was not greater for squamous cell tumors. The results of this study offer encouragement for further epidemiologic studies of the possible protective role of Vitamin A against cancer.